Automated Networks at the Service Edge
Computer networks have gained greater potential for automation. At the 2018 CENIC annual conference, Kireeti Kompella of Juniper Networks discussed the rationale behind establishing self-driving networks and how such networks could provide considerable progress in the areas of cybersecurity, user service, and skills advancement.
Kompella described a self-driving network as a data network whose operational processes have been automated. More specifically, it discovers its own components, creates the topology of the network, automatically comes up with basic network configuration, figures out what services are needed, connects people into the network, and monitors itself. Ideally, it can even predict where new capacity or bandwidth is needed.
These developments would allow operations staff to work at a higher level, freeing them to think more in terms of how a network can grow. Network security would also improve, as a machine can work more swiftly and efficiently than a person when diagnosing a problem.
“Events on the network don’t usually need immediate response,” said Kompella, “except when it comes to security. The faster you can detect, analyze, and remediate a breach in security, the better off you are. So I think that’s one place where having a machine behind the scenes figuring out what to do and doing it is really helpful.”
Kompella emphasized the importance of establishing self-driving networks at the service edge itself, where it will directly impact network users. This technology should, he argued, be a service made for and shaped by user needs.
“The challenge is to create a self-driving network at the edge of the network, where you offer services to your clients and can monitor those services and be proactive about them,” said Kompella. “The focus should be on making service better and more responsive.”
The complete vision, to Kompella, would involve designing a network that gathers user data and uses that data to shape itself accordingly. Once the service edge runs autonomously, self-driving technology can be extended to all parts of the network. This, Kompella said, would be the winning combination of data analytics and response automation that would define self-driving networks.
Like all advances in automation, self-driving networks would have a significant impact on the human workforce, especially with regard to which programming skill sets would become relevant and in demand.
“The impact of self-driving networks,” said Kompella, “is that you need a very different skill set. You do still need certified network engineers who know the ins and outs of [border gateway protocol] policy, but you now also need people who know artificial intelligence policy. You go from configuring a service to designing a service, from being reactive to proactive, from putting in the right access control list, to having the system say, ‘This is normal behavior’ or ’This is anomalous behavior.’ You need to think differently about building networks.”
At the close of his talk, Kompella called on all networking organizations to commit to groundbreaking new shifts in technology, identifying research and education (R&E) network organizations like CENIC as being leaders of networking innovation and change.
“I think we need a vision in networking that goes far beyond what we are doing today,” said Kompella. “We tend to be a little scared. But, I think R&E networks are a little different. You’re a little bolder, and you were much more out there when it came to things like OpenFlow and software-defined networking. We need a vision to aim toward, and we need to be really bold with it.”